Re: Caching Algorithms

From: Rodney Holm <rodneyh@dont-contact.us>
Date: Thu, 30 Jul 1998 13:43:15 -0600

Showing a page that the cache knows is incorrect does not seem to be a very
good idea. This could potentially lead to someone using the information on
the cached page as up-to-date. For instance on a stock quote site, the user
could choose to sell their stock based upon this incorrect quote. This is but
one example of how serving a stale cached copy could cause a serious problem.

If a site tells a cache that it doesn't want to be cached or tells a cache that
it wants to expire after a certain period of time, there is probably a good
reason for it.

Jordan Mendelson wrote:
>
> Curious, it seems that Squid, when receiving a web page with the no-cache or
> expires tags will not send it back to the client after it expires, but I was
> curious....
>
> If the site goes down and we have that cached copy in memory or on disk, why
> not send it back instead of the "read timeout" message. This happens (a lot)
> with http://slashdot.org. It seems to get overloaded a lot, the link goes
> down, etc.
>
> This is how Netscape's internal cache works (if it can't contact the host,
> it displays the previously cached copy even if the page instructed not to
> cache or to expire).
>
> Jordan
>
> --
> Jordan Mendelson : http://jordy.wserv.com
> Web Services, Inc. : http://www.wserv.com

-- 
-------------------------------------------------------------------------------
Rodney D. Holm                  rodneyh@apexxtech.com
Apexx Technology, Inc.          http://www.apexxtech.com
-------------------------------------------------------------------------------
Received on Thu Jul 30 1998 - 12:46:53 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:41:18 MST